Page 1
Task-switching and ERPs 1
This is the unedited authors’ version of a paper published elsewhere. This work should be cited as follows: Capizzi M., Feher K., Penolazzi B., Vallesi A. (in press). Task-‐switching preparation across semantic and spatial domains: an event-‐related potential study. Biological Psychology http://dx.doi.org/10.1016/j.biopsycho.2015.06.011. This version of the article may not exactly replicate the final version published in the Biological Psychology journal. It is not the copy of record.
Running Head: Task-switching and ERPs
Task-switching preparation across semantic and spatial domains: An event-related potential
study
Mariagrazia Capizzia,#, Kristoffer Fehérb, Barbara Penolazzic,d, Antonino Vallesia,e
aDepartment of Neuroscience, Università degli Studi di Padova, Padova, Italy
bDepartment of Psychiatric Neurophysiology, University Hospital of Psychiatry, University of Bern, Switzerland
cDepartment of General Psychology, Università degli Studi di Padova, Padova, Italy
dDepartment of Life Sciences, Università di Trieste, Trieste, Italy
eCentro di Neuroscienze Cognitive, Università degli Studi di Padova, Padova, Italy
#Corresponding Author’s address:
Mariagrazia Capizzi
Department of Neuroscience, Università degli Studi di Padova, Padova, Italy
Via Giustiniani, 5, 35128 Padova
Phone: +39 049 821 7181
E-mail: [email protected] or [email protected]
E-mails: Fehér K. ([email protected] ), Penolazzi B. ([email protected] ),
Vallesi A. ([email protected] )
Abstract
Page 2
Task-switching and ERPs 2
Previous event-related potential (ERP) studies have identified the specific electrophysiological
markers of advance preparation in cued task-switching paradigms. However, it is not completely
clear yet whether there is a single task-independent preparatory mechanism for task-switching or
whether preparation for a switch can be selectively influenced by the domain of the task to be
performed. To address this question, we employed a cued-task switching paradigm requiring
participants to repeat or to switch between a semantic and a spatial task. The behavioural results
showed a significant switch cost for both domains. The ERP findings, however, revealed that switch
and repeat trials for semantic and spatial domains differed in the amplitude modulation of an early
P2 and a sustained negativity both expressed over fronto-central scalp regions. Further differences
between the two domains also emerged over posterior-parietal electrodes. This pattern of data thus
shows that advance preparation in task-switching can be selectively modulated by the domain of the
task to be performed.
Keywords: ERPs; advance preparation; task-set; semantic processing; spatial processing.
Page 3
Task-switching and ERPs 3
1. Introduction
A hallmark of cognitive control is the ability to flexibly switch between tasks. One of the most
used tools to investigate such ability is the task-switching paradigm in which participants have to
repeat the same task or to switch between different ones. The general finding for task-switching
paradigms is that response time (RT) gets longer and accuracy decreases for switch trials as
compared to repeat trials, a phenomenon known as the “switch cost” (see Kiesel et al., 2010;
Monsell, 2003, for reviews). The switch cost is reduced but not completely eliminated even by
providing participants in advance with an explicit cue that instructs them to change task (i.e., the
cued task-switching paradigm; Meiran, 1996). The observation that a residual switch cost still
emerges with preparation intervals longer than 1 sec (Rogers & Monsell, 1995) suggests that
advance preparation cannot fully compensate for the behavioural cost of alternating between
different tasks (see Jamadar et al., 2010a).
Some theories explain the switch cost during the cued task-switching paradigm by assuming that
an active task-set reconfiguration process would be implemented for switch trials as compared to
repeat trials in order to prioritize the new task-set against the previous one (e.g., Rogers & Monsell,
1995). Such a reconfiguration process is supposed to be time-consuming and highly dependent on
executive control. Support for this claim comes from the finding of a reduction of the switch cost
when the cue-target interval is increased and more time can thus be devoted to advance preparation.
Alternatively, other researchers attribute the switch cost to priming or other memory interference
processes from the previous task-set that would not necessarily entail executive control (e.g.,
Allport, Styles, & Hsieh, 1994; Wylie & Allport, 2000). This idea is strengthened by the
observation that the switch cost is reduced with longer inter-trial intervals, which has been taken as
evidence that allowing ample time before the subsequent trial is presented favors the spontaneous
decay of the previous task-set interference. More recently, however, it is accepted that both
reconfiguration and interference processes would contribute to the switch cost (e.g.,
Vandierendonck, Liefooghe, & Verbruggen, 2010).
Page 4
Task-switching and ERPs 4
A number of studies using event-related potentials (ERPs) support the role of an active task-set
reconfiguration process taking place during the cue-target interval. The excellent high temporal
resolution of ERPs indeed allows researchers to track the time course of switch and repeat trials that
follow the presentation of the cue and to compare the neural activity associated with each task
condition. In such a way, it is possible to determine whether, and to what extent, switch and repeat
trials can be differentiated during the preparation interval that precedes task performance.
Two main ERP components have been often associated with task-switching effects during the
preparation interval: a sustained posterior positivity, sometimes termed “differential switch
positivity” or simply “switch positivity” (e.g., Jamadar et al., 2010a; Karayanidis et al., 2011),
emerging around 300-400 ms after cue onset, and a concurrent or later sustained frontal negativity
(e.g., Astle, Jackson, & Swainson, 2008; Lavric, Mizon, & Monsell, 2008). Both brain potentials
are typically larger for switch as compared to repeat trials (see De Baene & Brass, 2014;
Karayanidis et al., 2010, for reviews), although some studies also reported an enhanced frontal
negativity for repeat trials before target onset (e.g., Nicholson, Karayanidis, Poboka, Heathcote, &
Michie, 2005).
The switch positivity has been replicated across different studies and task manipulations (e.g.,
Kieffaber & Hetrick, 2005; Kopp, Lange, Howe, & Wessel, 2014; Li, Wang, Zhao, & Fogelson,
2012; Miniussi, Marzi, & Nobre, 2005; Nicholson et al., 2005; Rushworth, Passingham, & Nobre,
2002). A general consensus exists on the fact that this positivity would reflect anticipatory task-set
reconfiguration processes that would be especially related to switch trials. In support of this
interpretation, Karayanidis and colleagues (2011; see also Lavric, Mizon, & Monsell, 2008) found
faster switch responses to be associated with larger amplitude of the switch positivity as compared
to slower switch responses, suggesting that such a slow positivity is linked to “a switch-specific
reconfiguration process” (p. 567).
Unlike the switch positivity, the functional meaning of the frontal negativity appears more
controversial, perhaps due to the fact that this brain potential has been reported in fewer studies as
Page 5
Task-switching and ERPs 5
compared to the switch positivity. Furthermore, most of the studies observing the frontal negativity
have used a common average reference, which led to the suggestion that the frontal negativity and
the switch positivity could represent the negative and the positive components of a dipolar
distribution, respectively (see De Baene & Brass, 2014; Jamadar et al., 2010a; Karayanidis et al.,
2010; Lavric et al., 2008).
However, contrary to this claim, Astle, Jackson and Swainson (2008) found that the two brain
potentials, which were measured in the same time interval, could be dissociated in task-switching
paradigms that manipulated advance preparation of different response-sets. That is, whereas the
switch positivity was present for both overt and covert (i.e., mental counting) responses, the frontal
negativity was observed only when the task required an overt response. Moreover, in a study using
a go/no-go version of the task-switching paradigm (Astle, Jackson, & Swainson, 2006), it was
found that only the switch positivity was present following both go and no-go trials. By contrast,
there was no difference in the frontal negativity between switch and repeat trials after a no-go trial,
which suggested that this potential was sensitive to the fact that the response-set had been inhibited
in the previous trial and this effect carried over to the current trial. Taking the above studies into
account, a plausible explanation for the frontal negativity would be thus related to advance
preparation of overt response-set processes (see Karayanidis et al., 2010).
In addition to these sustained positive- and negative-going potentials, another reliable ERP
signature often reported in the task-switching literature is an early cue-locked fronto-central
positivity (P2), emerging approximately at 200 ms after cue onset, which is usually larger following
a switch cue relative to a repeat cue (e.g., Finke, Escera, & Barceló, 2012; Periáñez & Barceló,
2009; West, Langley, & Bailey, 2011). The enhanced P2 amplitude for switch trials has been
generally attributed to the functioning of an early task-set updating process that would rapidly
“detect” a relevant change in the task to be performed (see also De Baene & Brass, 2014).
To sum up, from this brief review of the main electrophysiological correlates of advance
preparation in cued task-switching paradigms, it seems clear that preparing for a switch as
Page 6
Task-switching and ERPs 6
compared to preparing for a repeat trial can differentially modulate some specific brain potentials
developing during the cue-target interval. Most of the previous task-switching studies have focused
on investigating which cognitive factors may influence the ERP markers of advance preparation.
Among others, it has been shown that the electrophysiological correlates of task-switching
preparation are sensitive to: 1) the amount of information conveyed by the cue (e.g., Karayanidis et
al., 2009; Nicholson et al., 2006), 2) the duration of cue-target and inter-trial intervals (e.g., Li et al.,
2012; Nicholson et al., 2005), 3) the specific requirements (go vs. no/go) for response selection
(e.g., Astle et al., 2006; Gajewski & Falkenstein, 2011; Jamadar et al., 2010b) and 4) the
participants’ performance (fast vs. slow switch responses) in switching between tasks (e.g.,
Karayanidis et al., 2011; Lavric et al., 2008).
Much less is known about the role played by the domain of the tasks that are manipulated in cued
task-switching paradigms. In other words, it is still unclear whether preparation for a task-switching
is accomplished by a single, task-independent, central mechanism or whether it relies on different
mechanisms according to the specific domain of the task to perform. Such a gap is mainly due to
the fact that previous ERP studies have usually focused on the contrast between switch vs. repeat
trials pooling over the tasks among which participants had to switch. This choice has been often
motivated by the finding of a null behavioural interaction between the requirements to switch/repeat
task and the specific task rules to be implemented, such that the ERP data have been averaged
across the different tasks in order to increase the signal-to-noise ratio of switch and repeat trials
(e.g., Goffaux et al. 2006; Karayanidis et al., 2009; Nicholson et al., 2006). As a consequence, it is
not completely clear to date whether task-switching preparation is domain-independent or rather it
is influenced by the domain of the task to be performed (see also Ravizza & Carter, 2008).
Among the few researchers who have investigated task-switching across different tasks, Hsieh
and Wu (2011; see also Hsieh, Wu, & Lin, 2014) compared the electrophysiological correlates of
advance preparation in task-switching between stimulus-dimensions vs. response-mappings. The
authors reported both common and distinct modulations of cue-locked ERPs associated with the
Page 7
Task-switching and ERPs 7
two task-switching types, which suggests the presence of both shared and unique mechanisms
underlying preparation to shift across different tasks.
An issue which is still poorly explored, however, is the comparison of task-switching between
tasks that are typically processed in distinct brain regions, like for instance semantic and spatial
tasks, which are known to be mainly processed on the left and right hemisphere, respectively (e.g.,
Corbetta & Shulman, 2011; Fairhall & Caramazza, 2013; Thompson-Schill et al., 1997). In the
present study, we asked whether there might be different preparatory mechanisms when shifting, on
a trial-by-trial basis, between tasks that require participants to make a spatial decision vs. tasks
requiring a semantic decision. To our knowledge, only a previous study by Miniussi, Marzi and
Nobre (2005) tackled a similar research question using a cued task-switching paradigm. In their
experiment a symbolic cue predicted, with 80% validity, the stimulus-dependent task to be
performed on any given trial: a lexical-decision task (i.e., to decide whether a letter string was a real
word or not) or an angle-decision one (i.e., to decide whether an angle was acute or obtuse). The
authors found a similar scalp distribution of the switch positivity for verbal and spatial tasks, which
pointed to the conclusion that task-switching preparation would draw at least on some common
task-independent processes. Nevertheless, another key finding in Miniussi and colleagues’ (2005)
study was that frontal and parietal modulations after a cue switch were larger in the verbal task as
compared to the spatial one. This result thus suggests that the domain of the task to be performed
may also influence general task-switching preparation processes.
To further explore the electrophysiological correlates of task-switching preparation across
different domains, in the present context we decided to use a cued task-switching paradigm that,
unlike Miniussi and colleagues’ (2005) study, implemented the same stimulus materials for both
semantic and spatial tasks. This was done to maximally strengthen task-set reconfiguration for the
two domains during the preparation interval, as the appearance of the same stimuli for both
semantic and spatial tasks reduced the possibility of an additional later task-set reconfiguration
process afforded by the identity of the stimuli itself. Moreover, using exactly the same stimuli,
Page 8
Task-switching and ERPs 8
while manipulating the cognitive operations underlying different domains, allowed minimizing the
influence that the type of material could also exert on the electrophysiological correlates of task-
switching preparation. Accordingly, we designed a semantic task and a spatial task in which the
stimuli consisted of several animal pictures belonging to two semantic categories (i.e., “preys” and
“predators”) and arranged in such a way to form different spatial diagonals. The participants’ task
was to identify the deviant set of animals (either “preys” or “predators”) or the deviant angle as
compared to two other sets of animals belonging to the same semantic category or to two other
equal spatial configurations, respectively. Each task was predicted with full certainty by two
auditory cues that were assigned to the semantic and spatial domains in a counterbalanced order
across participants.
While participants performed the task, continuous electroencephalographic (EEG) activity was
recorded. To determine whether task-switching preparation differs according to the domain of the
tasks being switched, we computed three cue-locked ERP components (P2, frontal negativity and
switch positivity) that, as detailed above, have been shown to be sensitive to the requirement to shift
task being usually larger for switch than for repeat trials. On the basis of prior cued task-switching
studies, we expected to observe a reliable RT switch cost for both semantic and spatial domains. As
regards the ERP data, two main predictions could be put forward. If task-switching preparation
relies on a common task-independent mechanism, we expected to replicate the finding of larger P2,
frontal negativity and switch positivity amplitudes for switch trials as compared to repeat trials in
both domains. Conversely, if updating and/or task-set reconfiguration processes would differ
between the two domains, thus selectively influencing task-switching preparation, a general
prediction would be that semantic and spatial rules should differentially modulate the amplitude
and/or time course of the electrophysiological indexes of switch and repeat trials within the cue-
target preparation interval.
2. Method
Page 9
Task-switching and ERPs 9
2.1. Participants. Twenty-two volunteers took part in the experiment in exchange for course
credits or cash payment of 20 euro. All participants gave informed consent prior to their inclusion in
the study. They reported to have normal or corrected-to-normal visual acuity and normal hearing.
None of the participants had any history of drug or alcohol abuse, or history of psychiatric,
neurological or other medical illness. The study was approved by the Bioethical Committee of the
Azienda Ospedaliera di Padova and was conducted according to the guidelines of the Declaration of
Helsinki. Data from two participants were discarded because of excessive noise in the EEG
recording. We also discarded data from two left-handed participants (average score: -92.5, range:
45-100, in the Edinburgh Handedness Inventory (Oldfield, 1971) as they displayed a different ERP
lateralization pattern as compared to the right-handed participants (average score: 90). The data
from the remaining eighteen participants (mean age: 26.4 years, range: 20-46 years, 13 females)
were used for both behavioural and ERP analyses.
2.2. Apparatus and stimuli. Stimulus presentation and response collection were controlled by E-
prime 2 software (Schneider, Eschman, & Zuccolotto, 2002) running on a personal computer
connected to a 19″ LCD monitor. This computer was interconnected to an Intel Core laptop
computer recording continuous EEG. The stimulus materials consisted of two auditory cue stimuli,
comprising a high pitch sound with a frequency of 1500 Hz and a low pitch sound with a frequency
of 200 Hz, and of eighteen visual target stimuli that depicted land-living mammals subdivided into
9 prey and 9 predator animals. Only four-legged animals were included. All animals unambiguously
faced right and were slightly tilted (i.e., at 15° in a clockwise manner). The animals were presented
into three white circles that were arranged in a row and displayed in the center of the screen against
a grey background. Each circle contained three identical animal pictures that were positioned in
such a way to form a diagonal, with one picture positioned in the center of the circle and the other
peripheral two displayed at a distance of 2.3 cm from the central image (see Figure 1).
Page 10
Task-switching and ERPs 10
----------------------------------------------------------
PLEASE INSERT FIGURE 1 ABOUT HERE
----------------------------------------------------------
The angles of the diagonal resulting from the arrangement of the three animal pictures varied
randomly between the values of 22.5°/202.5°, 45°/225°, 67.5°/247.5°, 112.5°/292.5°,135°/315° and
157.5°/337.5° from an imaginary horizontal line across the circle (the slashes mean that the angles
denote the same diagonal arrangement). The deviation from this angle could be ± 22.5° away from
the other two angles. This means, for instance, that the angle of 45° could be attained by adding
22.5° to 22.5° or by subtracting 22.5° from 67.5. Horizontal and vertical arrangements were not
included to avoid any pop-out effect of the deviant angle that could dramatically ease the spatial
task.
2.3. Procedure and Task. The task was a cued task-switching paradigm, in which an auditory
cue preceded each target presentation indicating the semantic task or the spatial task. For half of the
participants, the high pitch sound was associated with the semantic rule and the low pitch sound
with the spatial rule. For the other half of the participants, the reverse associations were used. A trial
started with the presentation of the auditory cue that was played for 300 ms via two loudspeakers
(Yamaha NX-50) located on both sides of the screen. The sound intensity was set at a comfortable
level (i.e., ¼ of the maximum volume) that was maintained constant for all the participants.
Following a fixed time interval of 1900 ms after the cue offset, the target stimuli were then
displayed for 2200 ms. Thus, there were 2200 ms from cue onset to target onset (see Androver-Roig
& Barceló, 2010, for a quite similar interval in a modified task-cueing version of the Winsconsin
Card Sorting Test). The employment of such a long cue-target interval was aimed at enabling
participants to fully develop advance preparation before target onset, thus avoiding overlapping
activity with subsequent target-related processing. Along the same line, we also decided not to vary
Page 11
Task-switching and ERPs 11
the duration of cue-target interval either between blocks of trials, since prior work has shown that
participants may rely on different preparatory strategies for short vs. long block intervals
(Nicholson et al., 2005), or within blocks of trials, not to add a further shift between short and long
intervals to our main task-switching manipulation and in order to avoid any variable foreperiod
effects (e.g., Vallesi, Lozano & Correa, 2013).
The participants’ task was to identify which of the three circles showed a deviant item with
respect to the other two according to a semantic or a spatial rule. In the semantic task, participants
had to identify the circle containing the deviant animal. For instance, if there were two circles
containing prey animals and only one circle with a predator, they had to indicate the circle
displaying the predator irrespective of the diagonal arrangement of the animals (see Figure 1 in
which the correct response is the circle on the right). By contrast, in the spatial task, participants had
to ignore the semantic category of the stimuli and focus on their spatial arrangement by indicating
the circle displaying a deviant angle as compared to the other two circles (in Figure 1 the correct
response corresponds to the circle on the left). For each task condition there was only one
univocally correct response, in the sense that the circle displaying the deviant angle could not also
contain the deviant animal or vice versa. Switch/repeat trials were equally probable and
administered randomly.
Participants responded by pressing the “j”, the “k” or the “l” keys on the computer keyboard with
the index, middle or ring finger of their right hand or, in different blocks, by pressing the “s”, the
“d” or the “f” keys with the ring, middle or index finger of their left hand, respectively. Each key
was spatially associated with each circle so that the first key was to be responded to if the deviant
circle was the one positioned on the left, the middle key if the deviant circle was the central one,
and the third key if the deviant circle was positioned on the right. The deadline for the response was
2500 ms after stimulus onset. Following a variable inter-trial interval ranging from 1000 to 1500
ms, the next trial began.
Page 12
Task-switching and ERPs 12
The experiment consisted of 8 blocks of 50 trials each. Four blocks required a left-hand response
and the remaining four a right-hand response. Half of the participants started with the left hand and
the other half with the right hand. Before the EEG session, participants were administered with both
a tutorial, which carefully explained the task, and a practice phase. At the beginning of the practice
phase, all the animal pictures were presented one by one with their name below each figure.
Participants were asked to mentally classify them as preys or predators and to let the experimenter
know in case of any doubt about the corresponding semantic category. Following this phase, they
were to perform each single-task separately before practicing the two tasks together. Half of the
participants started with the spatial task and the other half with the semantic task. Moreover, in
order to familiarize themselves with the stimulus-response mapping, they also had to change from
left to right hand, and vice versa, through the practice blocks. Specifically, half of the participants
used the left hand in the two single-task blocks and the right hand in the task-switching block. The
remaining participants started with the right hand in the single-task blocks and changed to the left
hand in the task-switching block.
Each single-task practice block consisted of 30 trials. At the end of each block, participants were
informed about their mean RTs and mean accuracy rates. If accuracy was below 66% after the first
block, subsequent mini blocks of 15 trials each were presented until participants managed to
perform the task above 66%. If accuracy was between 66% and 80%, participants could decide to
receive more training or not. If, instead, accuracy was above 80%, the program automatically
passed on the next test block. The same criteria applied to the task-switching practice, which
consisted of a first block of 40 trials and of subsequent mini blocks of 20 trials each administered
only if participants’ performance failed to reach the threshold of 66% of accuracy (or if participants
wished to practice more when accuracy was between 66% and 80%).
In the practice phase, participants received a feedback which varied according to their
performance after each trial (the Italian word for “Correct” displayed in blue or the Italian words for
“Incorrect” and “No response. Try to be faster” displayed in red for 1500 ms). In addition to the
Page 13
Task-switching and ERPs 13
practice phase, both at the beginning of the EEG session and after the first four blocks of
experimental trials (i.e., when the hand used to respond had to be changed), participants were
briefly presented with a short practice block of 8 trials. Like what stated above, they had to reach
the threshold of 66% of accuracy to start the proper experimental session. A short rest between
blocks of trials was allowed.
2.4. EEG recording. Participants were seated in front of the computer monitor and were
instructed to avoid eye blinks and movements during cue and stimulus presentation. The EEG was
recorded using BrainAmp amplifiers (Brain Products, Munich, Germany) from 64 Ag/AgCl
electrodes that were mounted on an elastic cap (EASYCAP GmbH, Germany) according to the
extended 10-20 system. Electrooculographic (EOG) activity was recorded with an electrode placed
under the left eye and was also monitored through the scalp electrodes positioned in the proximity
of both eyes. Impedances for each channel were measured and adjusted until they were kept below
10 kΩ before testing. All electrodes were referenced to FCz during the recording and were re-
referenced off-line to the average of all of the electrodes. An electrode positioned at AFz served as
the ground electrode. Raw data were band-pass filtered between 0.1-100 Hz and digitized at a
sampling rate of 500 Hz.
3. Analysis
3.1. Behavioural data analysis. Data from practice trials, the first trial of each block, trials with
errors and trials without responses were discarded from the RT analysis. Mean RTs for correct
responses and Accuracy (percentage of correct responses) were analyzed separately through
repeated-measures ANOVAs with Hand (left, right), Domain (semantic, spatial) and Switching
from the previous task (repeat, switch) as within-participants factors.
Page 14
Task-switching and ERPs 14
3.2. Electrophysiological data analysis. Signal pre-processing was performed using BrainVision
Analyzer 2.0 (Brain Products GmbH). Raw data were first band-pass filtered off-line with cutoffs at
0.1 and 30 Hz (Butterworth zero phase, 12 dB/Oct). An ocular correction algorithm based on
independent component analysis (ICA) was performed on the continuous data to correct for eye
movements and blink activity. Electrodes that were consistently bad during the entire recording
were replaced through spherical spline interpolation (Perrin, Pernier, Bertrand, & Echallier, 1989).
Overall, only four electrodes (T8, TP10, CPz and AF8) were interpolated for three participants (one
electrode each for two participants and two electrodes for another participant). The data were then
re-referenced to the average of all electrodes. They were finally segmented into epochs [-200, 2200
ms] with respect to the cue onset (baseline ±50 ms around the cue-onset; see Jamadar et al., 2010a).
Epochs were discarded if, on any channel, absolute difference between two sampling points
exceeded 30 µV/ms, if peak-to-peak deflections in a segment exceeded ±100 µV within intervals of
200 ms, if amplitude exceeded a value of ±80 µV and if activity was lower than 0.1 µV within
intervals of 200 ms. Finally, each epoch was visually inspected and trials containing any residual
artifact were manually removed. After artifact rejection, the total numbers of artifact-free trials per
condition were: 1394 for the semantic-repeat, 1286 for the semantic-switch, 1226 for the spatial-
repeat and 1243 for the spatial-switch condition. A minimum criterion of 28 artifact-free trials per
condition and participant was required to ensure a sufficient signal-to-noise ratio. Only trials with
correct behavioural responses were analysed. In addition, practice trials and the first trial of each
block were excluded from further analysis. Four separate grand average waveforms were
constructed relative to cue categories: semantic-repeat, semantic-switch, spatial-repeat and spatial-
switch, according to whether semantic and spatial trials were signaled either by the same cue (i.e.,
“semantic-semantic” or “spatial-spatial” trial sequences) or by a different cue with respect to that
used in the previous trial (i.e., “spatial-semantic” or “semantic-spatial” sequences).
The ERP analysis focused on the following cue-locked potentials: frontal P2, frontal negativity
and switch positivity that were chosen on the basis of visual inspection of the grand average
Page 15
Task-switching and ERPs 15
waveforms and according to prior literature. The P2 amplitude was analyzed over fronto-central
electrodes (left: F1, F3, FC1, FC3; midline: Fz, FCz; right: F2, F4, FC2, FC4) in a time window
ranging from 200 to 240 ms after cue onset (see Finke, Escera, & Barceló, 2012 and West, Langley,
& Bailey, 2011, for very similar analysis windows and electrodes). A repeated-measures ANOVA
tested for amplitude differences in the P2 with Domain (semantic, spatial), Switching from the
previous task (repeat, switch) and Electrode side (left, midline, right) as within-participant factors.
Regarding the analysis of the frontal negativity and the switch positivity, ERP amplitudes were
measured from fronto-central (left: F1, F3, FC1, FC3; midline: Fz, FCz; right: F2, F4, FC2, FC4)
and posterior-parietal (left: P1, P3, PO3, PO7; midline: Pz, POz; right: P2, P4, PO4, PO8)
electrodes, respectively, where the two brain potentials were maximally expressed (e.g.,
Karayanidis et al., 2009; West, Langley, & Bailey, 2011). As depicted in Figures 3 and 4, negative
and positive waveforms over fronto-central and posterior-parietal scalp regions almost overlapped
in time. Therefore, we analyzed the frontal negativity and the switch positivity within the same
latency range. Three time bins of 600 ms each were selected in order to explore the time course of
switch and repeat trials across the whole cue-target interval: (1) 400-1000 ms, (2) 1000-1600 ms,
and (3) 1600-2200 ms. Amplitude differences over fronto-central and posterior-parietal regions
were tested using a five-way repeated-measures ANOVA with the within-participant factors of
Scalp region (fronto-central, posterior-parietal), Time bin (1, 2, 3), Domain (semantic, spatial),
Switching from the previous task (repeat, switch) and Electrode side (left, midline, right). For the
scalp region factor, we pooled over all the above-mentioned electrodes that were contained within
each region. For all ERP analyses, amplitude was calculated as the mean voltage measured across
the pooled electrodes that were included in a particular montage (e.g., left electrodes side) and
within the specified temporal window. Significant effects of Electrode side were reported only if
they interacted with Domain, Switching or both. The Greenhouse-Geisser correction was applied
when sphericity assumption was violated according to the Mauchly’s test (Jennings & Wood 1976).
Corrected degrees of freedom and corrected probability values are reported.
Page 16
Task-switching and ERPs 16
For both ERP and behavioural analyses, post-hoc comparisons with the Tukey’s HSD test were
used to analyze both pair-wise comparisons within significant interactions and significant main
effects involving more than two levels.
4. Results
4.1. Behavioural results
4.1.1. RTs. The significant main effect of Switching [F(1,17)=20.55, p<.001, partial η2=.5]
showed that participants were slower when they had to switch from one task to another rather than
when the same task was repeated. The main effect of Domain was also significant [F(1,17)=34.31,
p<.0001, partial η2=.6] being RTs longer for the semantic domain than for the spatial domain. The
switch cost was not affected by the domain of the task to be performed as revealed by the non-
significant Domain x Switching interaction [F<.1] (see Figure 2.A).
----------------------------------------------------------
PLEASE INSERT FIGURE 2 ABOUT HERE
----------------------------------------------------------
Although there was no asymmetrical switch cost in our data, as indexed by the lack of a
significant Domain x Switching interaction (e.g., Martin et al., 2011), we calculated for both
domains an index of task-switching by computing the RT difference between switch and repeat
trials (62 ms for the semantic domain vs. 58 ms for the spatial domain) in order to directly compare
the magnitude of the switch cost across the two domains. A paired two-tailed t-test on these
behavioural indexes confirmed no difference in the switch cost between semantic and spatial
domains [t(17)=.21, p=.83]. The main effect of Hand and all the interactions involving Hand as a
factor were not significant (all ps>.1). Accordingly, in the EEG analysis the data across the two
hands were collapsed in order to increase the signal-to-noise ratio.
Page 17
Task-switching and ERPs 17
4.1.2. Accuracy. The main effect of Switching [F(1,17)=6.54, p=.02, partial η2=.2] mirrored the
RTs data by showing that participants were less accurate in the switch condition as compared to the
repeat condition. There was also a significant main effect of Domain [F(1,17)=4.62, p=.04, partial
η2=.2] indicating that accuracy was higher for the semantic domain than for the spatial one, a result
that goes in the opposite direction with respect to the RTs data showing a speed-accuracy trade off.
The interaction between Domain and Switching was far from significance [F(1,17)=2.06, p=.1,
partial η2=.1] (see Figure 2.B). None of the remaining terms of the ANOVA reached statistical
significance (all ps>.1).
4.2. Electrophysiological results
Cue-locked ERPs over fronto-central and posterior-parietal electrodes as a function of repeat and
switch trials are displayed separately for the semantic and the spatial domain in Figures 3 and 4,
respectively.
The ANOVA on the frontal P2 amplitude showed only a marginal significant Domain x Switching
interaction [F(1,17)=3.83, p=.06, partial η2=.1]. Post-hoc comparisons for this interaction revealed
that repeat spatial trials tended to elicit a larger P2 amplitude as compared to switch trials (p=.06),
whereas there was no difference between repeat and switch trials in the semantic domain (p=.9).
The ANOVA carried out on fronto-central and posterior-parietal regions showed both a
significant main effect of Scalp region [F(1,17)=7.82, p=.01, partial η2=.3], indexing a more
positive voltage over posterior-parietal electrodes than over fronto-central ones, and a significant
main effect of Time bin [F(2,34)=7.31, p=.002, partial η2=.3] with a more negative voltage in the
first time bin as compared to the second one (p=.001). There was no difference between the second
and third time bins as well as between the first and third time bins (ps>.1). Scalp region interacted
with Time bin [F(1.36, 23.12)=37.49, p<.001, partial η2=.6], with Time bin and Switching [F(1.29,
21.96)=11.05, p=.002, partial η2=.3] and with Domain and Switching [F(1,17)=28.59, p<.001,
partial η2=.6]. Time bin also interacted with Switching [F(2,34)=5.81, p=.006, partial η2=.2] and
Page 18
Task-switching and ERPs 18
with Domain and Switching [F(2,34)=3.39, p=.04, partial η2=.1]. These two-way and three-way
interactions were better qualified by a significant four-way interaction between Scalp, Time bin,
Domain and Switching [F(2,34)=6.06, p=.005, partial η2=.2]. Post-hoc comparisons for this four-
way interaction showed the following results.
In the first time bin (400-1000 ms), there was no difference for the semantic domain between
repeat and switch trials in either the fronto-central or the posterior-parietal scalp region (ps>.6).
Conversely, for the spatial domain switch trials in the first time bin were associated with a larger
positivity over posterior-parietal electrodes (p=.0002) and with a concomitant larger negativity over
fronto-central ones (p=.001).
In the second time bin (1000-1600 ms), post-hoc comparisons showed that whereas for the spatial
domain the frontal negativity was larger for switch trials as compared to repeat trials (p=.006), an
opposite pattern was found for the semantic domain with a larger frontal negativity for repeat trials
as compared to switch trials (p=.003). There was no difference between switch and repeat trials
over the posterior-parietal region for both semantic and spatial domains (all ps>.1)
In the third time bin (1600-2200 ms), post-hoc comparisons for the spatial domain confirmed a
larger frontal negativity for switch trials as compared to repeat trials (p=.02) and no difference
between the two trial types over the posterior-parietal region (p=.8). For the semantic domain, a
larger frontal negativity was instead associated with repeat trials as compared to switch trials
(p=.0001), whereas the opposite was observed for the posterior-parietal region with more negative
amplitude for switch trials as compared to repeat trials (p=.003).
There were no other significant main effects or interactions except for the Time bin x Switching x
Electrode side interaction [F(2.04, 34.76)=4.14, p=.02, partial η2=.1]. Post-hoc comparisons for this
interaction showed that in the first time bin switch trials were associated with a larger positive
amplitude as compared to repeat trials over the left electrode side (p=.001), whereas there was no
difference between switch and repeat trials in either the midline and the right side (ps>.1). In both
Page 19
Task-switching and ERPs 19
the second and the third time bin, switch trials did not differ from repeat trials in either of the three
electrode sides (all ps>.1).
------------------------------------------------------------------------
PLEASE INSERT FIGURES 3 AND 4 ABOUT HERE
------------------------------------------------------------------------
As already mentioned in the Behavioural results section, although the magnitude of the switch
cost did not differ between the two domains, the semantic task was associated with longer RTs as
compared to the spatial one. Accordingly, one might wonder whether the differences observed
between the two domains in the modulation of the P2, frontal negativity and switch positivity truly
reflected the involvement of different task-switching preparation processes or whether instead they
could be partly accounted for by the different task demands exerted by semantic and spatial
domains. In order to explore this possibility, we re-analysed the ERP data after having equated RTs
between the two domains. To do so, we excluded from the analysis of each domain a subset of the
fastest and slowest responses for both switch and repeat task conditions with the constraint,
however, that no more than 15 trials for each participant and condition should be rejected in order to
maintain an acceptable number of trials for the subsequent ERP analysis.
Behaviourally, this re-analysis confirmed a significant main effect of Switching [F(1,17)=14.89,
p=.001, partial η2=.4], with longer RTs when participants had to switch from one task to another as
compared to when they had to repeat the same task. The lack of a significant Domain x Switching
interaction [F(1,17)=0.43, p=.5, partial η2=.02] showed that there was no evidence for the switch
cost to be affected by the domain of the task to perform. As expected, after having equated RTs for
the two domains, the main effect of Domain was no longer significant [F(1,17)=1.3, p=.2, partial
η2=.07]. Crucially for our goal, the ERP statistical analysis of the trials equated for the two domains
replicated all the main results reported above.
Page 20
Task-switching and ERPs 20
5. Discussion
The main aim of the current study was to investigate whether advance task-switching preparation
could be differentially modulated according to the domain of the task to be performed, that is,
semantic or spatial. In order to strengthen task-set reconfiguration processes for the two domains
and minimize the influence that the type of material could have on the electrophysiological
correlates of advance preparation, we designed a novel paradigm that allowed us to administer the
same set of stimuli for both semantic and spatial domains.
The behavioural results replicated previous task-switching studies, being RTs longer and accuracy
lower for switch trials as compared to repeat trials. The data also showed participants to be slower
in the semantic task than in the spatial task in line with former work (e.g., Miniussi et al., 2005).
Importantly, however, the magnitude of the switch cost did not differ between semantic and spatial
domains and accuracy was higher for the semantic one. These results thus show that the main effect
under investigation in the present work, namely, the switch cost was similar for the two domains
despite the differences observed between semantic and spatial tasks in both RTs and accuracy data.
To summarize the ERP results, waveforms elicited by semantic and spatial domains showed
several differences that emerged as early as 200 ms after cue onset in the latency range of the
frontal P2. Whereas for the spatial domain the P2 amplitude tended to be larger for repeat trials as
compared to switch trials, for the semantic domain the P2 amplitude was not modulated by the
requirement to repeat or to switch task. Later on, the two domains differed reliably in the
modulation of both frontal and posterior brain potentials.
On the one hand, when participants switched from the semantic to the spatial domain, during the
400-1000 ms time bin switch spatial trials elicited both a larger positivity over posterior-parietal
electrodes and a concomitant larger negativity over fronto-central ones. From 1000 ms until the end
of the cue-target interval, spatial ERPs were only characterized by a more sustained negativity for
switch trials as compared to repeat trials over the fronto-central scalp region. On the other hand,
when participants switched from the spatial to the semantic domain, there was no difference
Page 21
Task-switching and ERPs 21
between switch and repeat trials during the first time bin (400-1000 ms) over either the fronto-
central or the posterior-parietal region. A later modulation within the semantic domain was found
over the fronto-central scalp region, with repeat trials being associated with larger negative
amplitude as compared to switch trials, in both the second and the third time bin. Such frontal
modulations were accompanied, in the third time bin (1600-2200 ms), by a more sustained
negativity for switch trials than for repeat trials over the posterior-parietal region.
The first difference between switch and repeat trials as a function of domain was already observed
in the time range of the frontal P2, which tended to be larger for repeat trials than for switch trials in
the spatial domain only. Therefore, the modulation of the P2 in the spatial domain differed from
previous results of enhanced P2 amplitude following a switch cue as compared to a repeat cue (e.g.,
Finke, Escera, & Barceló, 2012; Periáñez & Barceló, 2009; West, Langley, & Bailey, 2011). One
might interpret this pattern of data as reflecting an encoding benefit due to repetition of the same
cue. In the task-switching literature, there has been indeed a great deal of controversy regarding the
fact that employing a 1:1 mapping between cues and tasks may confound task-switch costs with
cue-switch costs (e.g., Logan & Bundesen, 2003). To overcome this problem, some studies have
used a 2:1 mapping between cues and tasks in such a way that a cue change could be also associated
with a task repetition (e.g., Hsieh and Wu, 2011). Although we acknowledge that in our design a
task change was always preceded by a cue change, this cannot account for our frontal P2 results in
the spatial domain. First, the association between cues and tasks was counterbalanced across
participants and the cues differed minimally at the physical level. Second, and more importantly, if
the modulation of the P2 observed in the spatial domain was related to cue repetition, we should
have expected to find the same pattern also for the semantic domain, which was not the case. Such
results thus challenge the idea that the cue-locked frontal P2 would merely detect a change in the
task to be performed since, as shown here, it was sensitive both to the requirement to repeat the
same task and to the domain of the task that needed to be repeated.
Page 22
Task-switching and ERPs 22
Another ERP deflection that was influenced by our task requirements was the switch positivity,
which was maximally expressed over the posterior-parietal scalp region in a time window ranging
from 400 to 1000 ms (see Figures 3 and 4, middle panel). Several studies have already highlighted
the importance of this brain potential in anticipatory preparation for a change in task (e.g.,
Karayanidis et al., 2010). Within the framework of the task-set reconfiguration theory, the presence
of a larger posterior positivity for switch trials vs. repeat trials would index the active
reconfiguration of the new task-set against the previous irrelevant one (e.g., Nicholson et al., 2005).
Hence, our finding of an enhanced positivity for switch spatial trials as compared to repeat ones fits
well with this account. However, it has been proposed that advance reconfiguration could also take
place on repeat trials, although to a lesser extent than what required in reconfiguring the new task-
set on switch trials. This would be further encouraged when switch and repeat trials have the same
probability of being presented within a block (e.g., Brass & von Cramon, 2004; Nicholson et al.,
2005). Supporting this proposal, repeat trials that are intermixed with switch trials have been found
to elicit a larger positivity as compared to all-repeat trials presented on single-task baseline blocks,
which suggests that some task-set reconfiguration processes could also occur on mixed repeat trials
(e.g., Wylie et al., 2009).
On the basis of the above-mentioned evidence, it might be reasonable to assume that in the current
study participants may have adopted different visual-attention strategies to accomplish the semantic
task with respect to the spatial one and that this could have enhanced reconfiguration processes on
both switch and repeat semantic trials (albeit note that this was not sufficient to eradicate the
behavioural switch cost in the semantic task). Namely, whereas on semantic trials it is likely that
participants adopted a speed-wise strategy more based on local processing in order to identify the
specific deviant animals among those included in the three circles, on spatial trials they had to scan
the animal pictures in a more global manner to pick up the deviant angle. This difference in strategy
formation would explain the finding of a more similar positive waveform for both switch and repeat
trials in the semantic domain vs. the spatial domain.
Page 23
Task-switching and ERPs 23
Alternatively, it could be possible to hypothesize that switch and repeat semantic trials both
elicited the same sustained positivity because of the greater difficulty of the semantic task, which
was indeed associated with longer RTs. However, this explanation does not hold to the extent that
the same ERP pattern was replicated even after having equated RTs for the two domains. Moreover,
the magnitude of the switch cost did not differ between the semantic and the spatial domain, which
rules out the possibility that the similar positivity observed for switch and repeat semantic trials
could be attributed to general task difficulty or to asymmetrical switch costs.
The employment of different strategies and the likely presence of different reconfiguration
processes for the two domains could have played a more critical role in the differential expression
of the switch positivity associated with semantic and spatial domains. Such cognitive factors may
also help explaining the different negative modulations and time courses that characterized the two
domains over the fronto-central scalp region. In all the three time bins considered for the ERP
analysis, switch spatial trials were more negative than repeat trials over fronto-central electrodes.
Conversely, in the semantic domain, within the first time bin there was no difference between
switch and repeat trials over the same fronto-central region. Afterwards, repeating the semantic task
gave rise to an increased negativity with respect to switching from the spatial to the semantic one, a
pattern that was significantly present during both the second and the third time bin (from 1000 to
2200 ms).
It is difficult to pinpoint the functional meaning of the differential modulation of the frontal
negativity by semantic and spatial domains shown here because, as outlined in the Introduction
section, no agreement has been reached yet on the role of this brain potential. Recently, it has been
suggested that the frontal negativity might reflect a general task preparation mechanism that would
not be specific for switch trials (e.g., Karayanidis et al., 2011). However, if we assume that in the
present context the frontal negativity was due exclusively to generic anticipation, then it would
make sense to predict a larger negativity for repeat trials than for switch trials in both domains,
which was not what we observed. In addition to the findings from the switch positivity, these results
Page 24
Task-switching and ERPs 24
thus point to the conclusion that the frontal negativity may also be affected in a different way by the
specific participants’ strategies activated during the preparation interval.
Our results for both the frontal negativity and the switch positivity differ from the study by
Miniussi and collaborators (2005) who found larger negative frontal and parietal modulations after
a cue switch to be associated with the more difficult verbal task. Several factors such as different
task requirements and timing parameters might have played a role in the differential outcome
between the two studies. Nevertheless, it should be emphasized here that asking participants to shift
across tasks that implemented exactly the same stimuli for both semantic and spatial domains
differentially influenced the specific ERP markers of task-switching preparation. Whilst this finding
points to the conclusion that task-switching preparation would draw on distinct task-dependent
mechanisms, it should be finally considered whether these results could be partially attributed to the
specific task domain transition employed in the current study. In other words, because our
participants had always to switch between two different domains (from semantic to spatial and vice
versa), one might argue that some carry-over interference effects would have come into play when
disengaging attention from one domain to the other, and that this eventually influenced task-
switching preparation processes. The same concern applies to Miniussi and colleagues’ (2005)
study, which also used a between-domain shift design, even if in their case this factor did not result
in dramatic differences between spatial and verbal domains in the expression of the ERP markers of
task-switching preparation. This finding thus suggests that our results cannot be explained solely by
the employment of a between-domain shift paradigm since, unlike our study, Miniussi and
colleagues (2005) showed common task-switching preparation mechanisms in the context of a
similar between-domain shift manipulation.
Along the same line, it should also be noted that the majority of previous ERP task-switching
studies have usually employed a switch “between” different tasks (i.e., letter task vs. digit task; e.g.,
Nicholson et al., 2005). Yet, since these studies typically collapsed the task factor, it is not possible
Page 25
Task-switching and ERPs 25
to get a complete picture of how the combination of task domain manipulation and between-domain
shift transition might affect task-switching preparation processes.
Nevertheless, another way of testing whether different types of task-switching would rely on
shared or distinct mechanisms is to manipulate task domain in a block-wise manner by keeping the
task domains among which participants have to switch separate across different blocks of trials
(e.g., Vallesi et al., 2015) or across different groups of participants (e.g., Hsieh & Wu, 2011). A
value of a within-domain shift design with respect to a between-domain shift one is that it allows
investigating task-switching preparation processes as a function of task domain by controlling for
possible carry-over interference effects. However, in the case of a between-participants study, a
drawback of using different groups of participants to compare distinct task-switching types is that it
is not possible to control for inter-subjects variability, which could also be a confounding variable.
Taking into account the above issues, perhaps a better manipulation to improve our understanding
of task-switching preparation as a function of task domain would be to orthogonally manipulate
task domain transition (within-domain shift and between-domain shift) in a full experimental design
and within the same individuals. Future studies should thus employ, within the same blocks, two
different tasks for each task domain (see Kieffaber and Hetrick, 2005, for a partial attempt in this
direction, with a design employing both within- and between-modality switches).
Finally, it should be acknowledged that some researchers recently begun to combine into the same
experiment single-shifts (i.e., a switch between stimulus dimensions: color or shape, or between
response effectors: hand or foot) and dual-shifts (i.e., a concurrent switch of both stimulus
dimensions and response effectors) in order to understand whether a dual-shift condition would be
associated with similar or distinct anticipatory processes as compared to a single-shift condition
(see Hsieh, Wu & Lin, 2014; Tieges et al., 2007; West, Bailey, & Langley, 2009, for ERP studies,
and Hübner et al., 2001; Philipp & Koch, 2010, for behavioural evidence). The main advantage of
using this kind of design is that it allows researchers to parametrically manipulate the “task shift
load” (i.e., single vs. dual-shift conditions; cf. Tieges et al., 2007) in order to explore how the ERP
Page 26
Task-switching and ERPs 26
markers of task-switching preparation would be modulated by the complexity of the switch
operation required on dual-shift vs. single-shift trials. However, when implementing this dual-shift
design in the context of a task domain manipulation, like the one used here, it is also important to
keep in mind that other factors such as higher memory load and concurrent reconfiguration of
multiple elements could indeed influence the genuine effects of task domain on task-switching
preparation processes.
To sum up, this brief review of the literature on different task-switching designs highlights the
importance of carefully selecting, according to the specific task goal of the study, the experimental
design that is more suitable to investigate task-switching ability. Here, we showed that a between-
domain shift transition across semantic and spatial tasks selectively influenced the specific ERP
signatures of task-switching preparation. Our results thus suggest that when participants have to
shift on a trial-by-trial basis between two tasks belonging to two separate cognitive domains, task-
switching preparation would rely on distinct mechanisms. Future studies will clarify whether it is
possible to generalize these conclusions to experimental settings in which task domain is
manipulated, for instance, in a block-wise manner.
Page 27
Task-switching and ERPs 27
Acknowledgments
MC and AV are funded by the European Research Council Starting grant n° 313692 (FP7/2007-
2013) to AV. The authors wish to thank Giovanni Galfano for his support in participants’
recruitment, Ettore Ambrosini for his help with statistical matters and Sandra Arbula for her
assistance with data collection. Thanks go also to Città della Speranza, Padova, for its invaluable
logistic support.
Page 28
Task-switching and ERPs 28
References
Allport, A., Styles, E. A., & Hsieh, S. (1994). Shifting intentional set: Exploring the dynamic
control of tasks. In C. Umilta & M. Moscovitch (Eds.), Conscious and nonconscious information
processing: Attention and performance XV (pp. 421–452). Cambridge, MA: MIT Press.
Androver-Roig, D., & Barceló, F. (2010). Individual differences in aging and cognitive control
modulate the neural indexes of context updating and maintenance during task switching. Cortex, 46
(4), 434-50.
Astle, D. E., Jackson, G. M., & Swainson, R. (2006). Dissociating neural indices of dynamic
cognitive control in advance task-set preparation: an ERP study of task switching. Brain Research,
1125, 94–103.
Astle, D. E., Jackson, G. M., & Swainson, R. (2008). Fractionating the cognitive control required
to bring about a change in task: a dense-sensor event-related potential study. Journal of Cognitive
Neuroscience, 20, 255–267.
Barceló, F., Escera, C., Corral, M. J., & Periáñez, J. A. (2006). Task switching and novelty
processing activate a common neural network for cognitive control. Journal of Cognitive
Neuroscience, 18 (10), 1734-48.
Brass, M., & von Cramon, D. Y. (2004). Decomposing components of task preparation with
functional magnetic resonance imaging. Journal of Cognitive Neuroscience, 16, 609-620.
Corbetta, M., & Shulman, G.L. (2011). Spatial neglect and attention networks. Annu. Rev.
Neurosci. 34, 569-599.
De Baene, W., & Brass, M. (2014). Dissociating strategy-dependent and independent components
in task preparation. Neuropsychologia, 62, 331-340.
Fairhall, S.L., & Caramazza, A. (2013). Brain regions that represent amodal conceptual
knowledge. Journal of Neuroscience, 33, 10552-10558.
Page 29
Task-switching and ERPs 29
Finke, M., Escera, C., & Barceló, F. (2012). The effects of foreknowledge and task-set shifting as
mirrored in cue- and target-locked event-related potentials. PLoS One, 7(11):e49486. doi:
10.1371/journal.pone.0049486.
Gajewski, P.D., & Falkenstein, M. (2011). Diversity of the P3 in the task-switching paradigm.
Brain Research, 1411, 87-97.
Goffaux, P., Phillips, N.A., Sinai, M., & Pushkar, D. (2006). Behavioural and electrophysiological
measures of task switching during single and mixed-task conditions. Biological Psychology, 72,
278–290.
Hsieh, S., & Wu, M. (2011). Electrophysiological correlates of preparation and implementation
for different types of task shifts. Brain Research, 1423, 41-52.
Hsieh, S., Wu, M., & Lin, F. (2014). Neural correlates of response-effector switching using event-
related potentials. Biological Psychology, 103, 332-348.
Hübner, R., Futterer, T., & Steinhauser, M. (2001). On attentional control as source of residual
shift costs: Evidence from two-component task shifts. Journal of Experimental Psychology:
Learning, Memory, and Cognition, 27, 640-653.
Jamadar, S., Hughes, M., Fulham, W.R., Michie, P.T., & Karayanidis, F. (2010a). The spatial and
temporal dynamics of anticipatory preparation and response inhibition in task-switching.
NeuroImage, 51, 432-49.
Jamadar, S., Michie, P.T., & Karayanidis, F. (2010b). Sequence effects in cued task-switching
modulate response preparedness and repetition priming processes. Psychophysiology, 47, 365-386.
Jennings, J. R., & Wood, C.C. (1976). The e-adjustment procedure for repeated-measures
analyses of variance. Psychophysiology, 13, 277–278.
Karayanidis, F., Mansfield, E. L., Galloway, K. L., Smith, J. L., Provost, A., & Heathcote, A.
(2009). Anticipatory reconfiguration elicited by fully and partially informative cues that validly
predict a switch in task. Cognitive, Affective, & Behavioral Neuroscience 9, 202–215.
Page 30
Task-switching and ERPs 30
Karayanidis, F., Jamadar, S., Ruge, H., Phillips, N., Heathcote, A., & Forstmann, B.U. (2010).
Advance preparation in task-switching: converging evidence from behavioral, brain activation and
model-based approaches. Frontiers in Psychology, 1:25.
Karayanidis, F., Provost, A., Brown, S., Paton, B., & Heathcote, A. (2011). Switch-specific and
general preparation map onto different ERP components in a task-switching paradigm.
Psychophysiology, 48 (4), 559-68.
Kieffaber,P.D., Hetrick, W.P. (2005) Event-related potential correlates of task switching and
switch costs. Psychophysiology, 42, 56–71.
Kiesel, A., Steinhauser, M., Wendt, M., Falkenstein, M., Jost, K., Philipp, A., and Koch, I. (2010).
Control and interference in task-switching – A review. Psychological Bulletin, 136, 849-874.
Kopp, B., Lange, F., Howe, J., & Wessel, K. (2014). Age-related changes in neural recruitment
for cognitive control. Brain and Cognition, 85, 209-219.
Lavric, A., Mizon, G., & Monsell, S. (2008). Neurophysiological signature of effective
anticipatory task-set control: a task-switching investigation. European Journal of Neuroscience, 28,
1016–1029.
Li, L., Wang, M., Zhao, Q-J., & Fogelson, N. (2012). Neural mechanisms underlying the cost of
task switching: An ERP study. PloS One 7:e42233.
Logan, G. D., & Bundesen, C. (2003). Clever homunculus: Is there an endogenous act of control
in the explicit task-cuing procedure? Journal of Experimental Psychology: Human Perception and
Performance, 29, 575-599.
Martin, C.D., Barceló, F., Hernandez, M., & Costa, A. (2011). The time course of the
asymmetrical “local” switch cost: Evidence from event-related potentials. Biological Psychology,
86, 210-218.
Meiran, N. (1996). Reconfiguration of processing mode prior to task performance. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 22, 1423–1442.
Page 31
Task-switching and ERPs 31
Miniussi, C., Marzi, C.A., & Nobre, A.C. (2005). Modulation of brain activity by selective task
sets observed using event-related potentials. Neuropsychologia, 43, 1514-1528.
Monsell, S. (2003). Task switching. Trends in Cognitive Sciences, 7, 134–140.
Nicholson, R., Karayanidis, F., Davies, A., & Michie, P. T. (2006). Components of task set
reconfiguration: Differential effects of “switch-to” and “switch-away” cues. Brain Research, 1121,
160-176.
Nicholson, R., Karayanidis, F., Poboka, D., Heathcote, A., & Michie, P.T. (2005).
Electrophysiological correlates of anticipatory task-switching processes. Psychophysiology, 42 (5),
540–554.
Oldfield, R. C. (1971). The assessment and analysis of handedness: The Edinburgh inventory.
Neuropsychologia, 9(1), 97–113.
Periáñez, J.A., & Barceló, F. (2009). Updating sensory versus task representations during task-
switching: Insights from cognitive brain potentials in humans. Neuropsychologia, 47, 1160-1172.
Perrin, F., Pernier, J., Bertrand, O., & Echallier, J.F. (1989). Spherical splines for scalp potential
and current density mapping. Electroencephalogr Clin Neurophysiol, 72(2), 184-7.
Philipp, A.M., & Koch, I. (2010). The integration of task-set components into cognitive task
representations. Psychologica Belgica, 50, 383-411.
Ravizza, S. M., & Carter, C. S. (2008). Shifting set about task switching: Behavioural and neural
evidence for distinct forms of cognitive flexibility. Neuropsychologia, 46, 2924-2935.
Rogers, R. D., & Monsell, S. (1995). Costs of a predictable switch between simple cognitive
tasks. Journal of Experimental Psychology: General,124, 207–231.
Rushworth, M.F.S., Passingham, R., & Nobre, A.C. (2002) Components of switching intentional
set. Journal of Cognitive Neuroscience, 14, 1139–1150.
Schneider, W., Eschman, A., & Zuccolotto, A. (2002). E-Prime user's guide. Pittsburgh:
Psychology Software Tools Inc.
Page 32
Task-switching and ERPs 32
Thompson-Schill, S.L., D’Esposito, M., Aguirre, G.K., & Farah, M.J. (1997). Role of left inferior
prefrontal cortex in retrieval of semantic knowledge: a reevaluation. Proc. Natl. Acad. Sci. U. S. A.,
94, 14792-14797.
Tieges, Z., Snel, J., Kok, A., Plat, N., & Ridderinkhof, R. (2007). Effects of caffeine on
anticipatory control processes: Evidence from a cued task-switch paradigm. Psychophysiology, 44,
561-578.
Vallesi, A., Arbula, S., Capizzi, M., Causin, F., & D’Avella, D. (2015). Domain-independent
neural underpinning of task-switching: An fMRI investigation. Cortex, 65, 173-183.
Vallesi, A., Lozano, V.N., & Correa, A. (2013). Dissociating temporal preparation processes as a
function of the inter-trial interval duration. Cognition, 127, 22-30.
Vandierendonck, A., Liefooghe, B., & Verbruggen, F. (2010). Task switching: Interplay of
reconfiguration and interference. Psychological Bullettin, 136, 601-626.
West, R., Bailey, K., & Langley, M.M. (2009). An investigation of the neural correlates of
attention and effector switching using ERPs. Cognitive, Affective, and Behavioral Neuroscience, 9,
190-201.
West, R., Langley, M.M., & Bailey, K. (2011). Signaling a switch: Neural correlates of task
switching guided by task cues and transition cues. Psychophysiology, 48, 612-623.
Wylie, G., & Allport, A. (2000). Task switching and the measurement of ‘switch costs’.
Psychological Research, 63, 212-233.
Wylie, G. R., Murray, M. M., Javitt, D. C., & Foxe, J. J. (2009). Distinct neurophysiological
mechanisms mediate mixing costs and switch costs. Journal of Cognitive Neuroscience, 21, 105-
118.
Page 33
Task-switching and ERPs 33
Figure captions
Figure 1. Example of stimulus material. When the task required a semantic decision, participants
had to choose the circle containing the deviant animal (i.e., prey or predator) as compared to the
other two circles. In the figure, the circle on the right was the target since it contained a predator
(i.e., tiger), whereas the other two displayed prey animals (i.e., deer and zebra). By contrast, in
case of a spatial decision, the correct response would have been the circle on the left since the
arrangement of the animal pictures created a deviant angle as compared to the other two circles.
Page 34
Task-switching and ERPs 34
Figure 2. (A). Mean response times (RTs) and (B) percentage of correct responses as a function of
Domain (semantic, spatial) and Switching from the previous task (repeat, switch). Vertical bars
represent standard errors of the mean.
Page 35
Task-switching and ERPs 35
Figure 3. (A) Cue-locked grand averages of the semantic ERP waveforms recorded at pooled
fronto-central electrodes (top panel) and posterior-parietal electrodes (middle panel) as a function
of Switching from the previous task (repeat, switch) and Electrode side (left, midline, right). ERPs
of interest (P2, frontal negativity and switch positivity) are marked on midline electrodes only for
general visualization. (B) Differences in the ERP topography between switch and repeat trials for
the semantic domain. The difference maps (switch minus repeat) are shown for the time bins used
for the ERP analysis of the P2, frontal negativity and switch positivity.
Page 36
Task-switching and ERPs 36
Figure 4. Cue-locked grand averages of the spatial ERP waveforms recorded at pooled fronto-
central electrodes (top panel) and posterior-parietal electrodes (middle panel) as a function of
Switching from the previous task (repeat, switch) and Electrode side (left, midline, right). ERPs of
interest (P2, frontal negativity and switch positivity) are marked on midline electrodes only for
general visualization. (B) Differences in the ERP topography between switch and repeat trials for
the spatial domain. The difference maps (switch minus repeat) are shown for the time bins used for
the ERP analysis of the P2, frontal negativity and switch positivity.
Page 37
Task-switching and ERPs 37